Unsteady Heat Transfer in Laminar Non-

Newtonian Boundary Layer over a Wedge

The transient thermal response of a power-law type non-Newtonian, laminar RAMA SUBBA REDDY GORLA
boundary layer flow over a wedge is investigated. Consideration is given to the
Department of Mechanical Engineering
case of a step change in surface temperature. Details of the transient heat flux and
Cleveland State Universlty
the temperature field are obtained and have been presented graphically. The range
of Prandtl numbers investigated is from 5 to 1000 while the viscosity index was Cleveland, OH 44115
allowed to vary from 0.1 to 5.0.

SCOPE
I t is often necessary to compute the unsteady, forced, con-
vective heat transfer from a body whose surface temperature
is changing with time. Although the published literature con-
tains several investigations dealing with the steady state heat
transfer in nowNewtonian fluids, analyses for the unsteady heat
transfer problem are scarce.
The present work is undertaken in order to investigate the
unsteady heat transfer in non-Newtonian fluid boundary layer
flow over a wedge. The governing boundary layer equations are
formulated and converted into ordinary differential equations
by appropriate transformations. The solution then is obtained
by first transforming the energy equation in the Laplace
transform variable and seeking a series expansion. The series
solution is constructed so that it is valid for small as well as large
times.

CONCLUSIONS AND SIGNIFICANCE

I n this paper, an analysis is presented concerning the un-
steady heat transfer in a non-Newtonian, laminar boundary
layer flow over a wedge. The thermal boundary layer thickness
is seen to increase with time as a result of the step change in
surface temperature and ultimately coincide with the steady
state distribution as time tends to infinity. The transient heat
flux increases with increasing Prandtl number and approaches
the steady state value for very large times. The time required
for reaching the steady state condition may be calculated from
the results presented in this paper.
The step function results presented here serve as a funda-
mental solution, since by a superposition technique these results
may be generalized to apply for arbitrary time variation in the
surface temperature.

0001-1541-82-5123-06$2.00
0 The American Institute of Chemical Engineers, 1982

INTRODUCTION the aerodynamics of the modern high speed vehicles should be

Even though a lot of research work has been done in the general
considerations of linear, Newtonian relationship between the fluid
stress and rate of strain, it has been relatively recently that much
research is being done on many aspects of the mechanics of fluids
possessing non-linear, non-Newtonian relationships between stress
and strain. There exists a need for systematic inquiry of funda-
mental nature into many aspects of non-Newtonian fluids. Fluids
such as molten plastics, pulps, slumes, emulsions, etc., are examples
of non-Newtonian fluids and are found to have several applications
in the chemical, food, construction, petroleum production and
power engineering industries.
Non-Newtonian boundary layer theory is relevant to a wide
diversity of engineering activities, among which may be cited the
possibility of reducing frictional drag on bearings and on immersed
bodies such as ship hulls and submarines. Boundary layer devel-
opment determines the transport phenomena associated with
non-Newtonian flow in the entrance regions of tubes and ducts and
flow along banks of tubes in the shell side of heat exchangers. Drag
on centrifugal pump blades, free convection processes and the
development of the entrance region in non-Newtonian thin film
reactors and falling film evaporators are determined by boundary
layer characteristics. As has been pointed out by Luikov (1969),
based upon the rheological concepts since atmospheric air at alti-
tudes above 9 km has distinct viscoelastic properties.
In the 195Os, research efforts in non-Newtonian fluids were
concentrated on internal flow problems. A review of the subject
was given by Metzner (1965). Acrivos et al. (1960) and Shah et al.
(1962) were probably the first to study laminar boundary layer
flows of non-Newtonian fluids past external surfaces. Wolf and
Szewczyk (1966)analyzed the heat heat transfer to an incom-
pressible, laminar flow of power-law, non-Newtonian fluid from
an arbitrary symmetrical cylinder. Lee and Ames (1966) found
similarity solutions for forced convective heat transfer to power-law
fluids in the case of wedge-type flows.
All the previously mentioned investigationsdeal with steady state
heat transfer in non-Newtonian fluids. It is often necessary to
compute the unsteady forced convective heat transfer from a body
whose surface temperature is changing with time. Neither analyses
nor experiments for this condition have been reported in the lit-
erature so far.
The present work aims at investigating the unsteady heat transfer
in laminar, power-law type, non-Newtonian boundary layer flows
over wedges. The transient response behavior and the details of the
transient temperature fields for the case of step change in surface
temperature are obtained for various Prandtl numbers and a range

Page 56 January, 1982 AlChE Journal (Vol. 28, No. 1)

of values of n, the viscosity index. The unsteady stagnation point
heat transfer in non-Newtonian fluids has been studied by Gorla
(1980).

GOVERNING EOUATIONS

Let us consider a steady, incompressible, laminar and two-

dimensional boundary layer flow of a power law-type, non-New-
tonian fluid over a wedge of included angle ap, placed symmet-
rically in a uniform main stream. We assume the flow to be steady
and the free stream temperature to be constant at T m . Initially the
surface temperature is T m and at time t = 0, a step change in the
surface temperature is applied. The coordinate system along with
the flow model has been shown in Figure 1.
Assuming constant fluid properties and negligible viscous dis-
sipation, the governing equations under boundary layer approxi-
mations can be written as:
Figure 1. Coordinate system and flow development.
dU* dV*
Mass: -
ax*
+-
by*
=0
d
Momentum: u* -+ V* -
dU* dU* --cf”)” + ff” =P[Cf‘)2-1] (9)
dv
dX* dY*
where
- m(n 1) +
dT dT bT d2T ’=2mn-m+ 1
Energy:-+ u*-+ V*-=a.- (3)
dt* dX* ay* dy*2 The transformed boundary conditions are

For Power Law Fluids, 7,y = K

(3
-
Here, we note that U*(r*)is the velocity outside the boundary layer
(4)
f(0) = f’(0) = 0 and f’( m) =1
By means of a numerical solution of Eq.9, we obtainf”(0) = a2 for
(10)

various values of m and n. Primes in the above equations designate

and is equal to CX*m where C is a constant and m = p/(2 - p). differentiation with respect to v only.
The boundary conditions for the velocity field are: To transform the energy equation, we define
u*(x*,O) = V*(x*,O)= 0 and u * ( x * , m ) = U* (5) O=- T - T ,
The initial and boundary conditions for the temperature field T, - T ,
[’-
are:
T(x*,y*,O)= T m , T ( x * , O , t * ) = T m
+ (Tw - T m ) l(t*) and T(X*,m,t*) = T m (6)
=
n + l I
+ 2mn 2/3(1+n).t .
r(6m-2n-4)/3(n+ 1)

where l(t*) = 0 for t* < 0

= 1 for t* 1 0 The transformed energy equation may be written as
+
Pr[1- %z(n 2 - 3m)
COORDINATE TRANSFORMATION 3(n 1)+
We now define the dimensionless variables

x =-
X*
y = Y*. Rel/n+l The initial and boundary conditions are
L L
t*U, U* 6([,0)= 0,d(0,7) = l ( 7 ) and 6(-,7) = 0 (13)
t=- u=-
L u,
-.U ,
V = V* Rel/n+l u =-u* SOLUTION
ur
n

pU%-”L“ Defining the Laplace transform of 6 as

Re = ~

K
Defining a stream function %‘such that u = d%”/dyand v =
-d%’/dx the continuity is automatically satisfied. We may and applying the Laplace transformation to Eqs. 12 and 13 we
write have
= pf’(v)
.
l / ( l + n ) .,(zmn-n-mj/(n+l). G(v)
l - m + 2mn where
2m -nm - 1 aZ(3m 2n 1) + +* E2
).(vy-i)+mf
6(n 1)
= +
.y . ,(zm-mn-l)/(n+l) %2(n +
2 - 3m)
.E
(8) 32(E) = 3(n 1) +
It may be verified that the momentum equation becomes The appropriate boundary conditions are given by

AlChE Journal (Vol. 28, No. 1) January, 1982 Page 57

 TABLE1. VALUES OF FOR VARIOUS VALUES OF n AND Pr

We now seek a solution of the form n Pr = 5 P r = 10 Pr = 5 0

AP r = 100 Pr =500 Pr = loo0
0.1 1.6947 1.3451 0.7866 0.6243 0.3651 0.2897
0.3 0.9228 0.7324 0.4283 0.3399 0.1988 0.1578
0.5 0.7582 0.6018 0.3519 0.2793 0.1633 0.1296
where 1.0 0.6515 0.5171 0.3024 0.2400 0.1404 0.1114
3.0 0.6273 0.4979 0.2912 0.2311 0.1351 0.1073
5.0 0.6424 0.5098 0.2982 0.2366 0.1384 0.1098

It may be noticed that X is a function of [ which is still unknown.

We set u,(O) = 1 and uj(0) = 0 for all j 2 1 so that ON)= l / p is with boundary conditions
identically satisifed. Examining Eq. 17, we see that O ( - ) = 0 is
automatically satisifed. Substituting Eq. 17 into 15 and equating O,(O) = 1 and &(a)= 0 (24)
+
coefficients of like powers of (p A) one obtains a set of ordinary We now obtain a closed form solution for 0, as
differential equations from which closed form solutions for u&)
may be obtained. These details are not shown in the interest of
conserving space. O,(cy) =1 -- (3
It has been found that the first five terms in the series give sat-
isfactory convergence.
Taking the inverse of Eq. 17 we have for the transient temper-
ature field
(3
where

r&)= Incomplete Gamma Function

By differentiating Eq. 25 and then transforming to the [-system,
where we obtain

By means of Eqs. 20,22,25 and 26, the numerical values of A([)

for i 2 2 we have have been obtained through an iterative computer program. A
knowledge of the numerical results for X is essential in obtaining
the transient temperature distribution and surface heat flux.

RESULTS AND DISCUSSION

When 4 = 0, the series in Eq. 20 becomes identically unity and

thus cannot be used to determine &,. The value of A, was evalu-
h3 = (exp[(PrX)l/z[ - AT]] ated from Eq. 22 using the values of O',(O) obtained from Eq. 26.
The values of L have been tabulated in Table 1 for a wide range
of values of n and Pr. Figure 2 illustratesthe growth of the thermal

-
layer with time for n = 1. For the sake of brevity, the value of Pr
One may notice that H , varies from 0 to 1 as 7 varies from 0 to m = 10 and m = 1/3 (corresponding to a right-angle wedge, (3 = l/2)
and also as [ varies from 0 to a.By letting T QO in Eq. 18, we

obtain the steady state temperature distribution

n- 1.0

1.00 L Pr-10

We now obtain an expression for @(O,T) after differentiating Eq.

18 with respect to [ and substituting [ = 0 as
- -
e'(0,~)= -(Pr)1/2 (xT)-'/~ exp(-XwT)
-
- (Pr Xw)1/2 erf(XwT)l/2
+ 2 u;(o)- (prXw)-j/2. [rAw7(j/2)/r(j/2)](21)
By letting 7 - j=1

m, we obtain the steady state value for the wall

temperature gradient as
emrn)= e:(o) = -(PrL)1/2 + 2 u;(o). [ ~ r ~ , l - j / z (22)
j=l

A, will have to be evaluated from the steady state solution. The

steady state problem is described by the following equation
a2(3m 2n 1)
6(n 1)
++ + t2] - 3
dBs
=0 (23)
0 0.5
c
1.0 15

Flgure 2. Thermal boundary layer growth for n = 1 and Pr = 10.

Page 58 January, 1982 AlChE Journal (Vol. 28, No. 1)

have been chosen. The local wall heat flux may be written by to be a strong function of the value of the vixosity index. As the
Fourier's law as value of n increases, the thermal boundary layer thickness is found
to increase. The time required to reach the steady state conditions
increases with Prandtl number for a given value of n. The time
required for approaching steady state conditions may be calculated
from the results presented in this paper. As an example, let us
consider the boundary layer flow of carbopol940 solution (0.058%
solution) over a flat plate of 1-m length. At a mean temperature
(27) of 90°F (32"C), we have K = 0.0715 and n = 0.675. If the undis-
turbed free-stream velocity U , is assumed to be 0.1 m/s, the time
The local heat transfer coefficient is given by required to approach steady state conditions ( t i ) is found to be 85
seconds at r * = 0.5 m. For x* = 1 m, we find that t i = 123sec-
onds.

CONCLUDING REMARKS

The local Nusselt number therefore may be written as In this paper, we have presented an analysis investigating the
unsteady heat transfer characteristics of a non-Newtonian, laminar
Nur* = (hx*/k) boundary layer flow over a wedge. The transient response behavior
=-( .$rn+2n+ 1/3(n+1)
due to a step change in surface temperature has been considered.
In a situation where an arbitrary surface temperature variation is
n + l
imposed, the heat transfer characteristics can be deduced from the
- Rel/"+l - e l ( 0 , ~ )
By letting 7 -
in Eq. 28, one obtains the steady state Nusselt
number Nu,. From Eqs. 26 and 28, we have
(29) results for the case of a step change in surface temperature. Toward
this end, numerical superposition techniques may be used conve-
niently.

In the case of a Newtonian fluid (n = l ) , we obtain for the local ACKNOWLEDGMENTS

Nusselt number under steady state conditions:
+
Nu, = 0 . 4 3 5 8 ~ 2 ~ / ~ ( m
1)1/2Re1/2Pr1/3 (31)
The results for the steady state conditions are in excellent agree-
ment with published results. The ratio of the instantaneousto steady
state values of the heat flux are important for practical applications.
We may write
The ratio of Nu/Nu, can be evaluated by means of Eqs. 21 and 26
for any value of time with Prandtl number and the viscosity index
as parameters. The transient heat flux distribution as described by
Eq. 32 is illustrated by means of Figure 3 for several values of Pr
while n and m were set to be equal to 1 and '/3 respectively.
From the results obtained, we notice that the thermal boundary
layer thickness increases with time following a step change in
A portion of this work was performed while the author worked
as a summer Research Fellow at NASA Lewis Research Center in
Cleveland, Ohio. The author wishes to thank Dr. Robert W. Gra-
ham, Head of the Fundamental Heat Transfer Section at NASA
Lewis Research Center for his interest and encouragement. The
constructive remarks made by Professor David Wollersheim and
the anonymous referees have considerably helped to improve the
quality of the paper. The help offered by Dr.Robert Kadlec during
the process of revision of the paper to confirm with AIChE format
is appreciated.
NOTATION
f = nondimensional stream function
h = heat transfer coefficient
K = viscosity coefficient in Ostwald-de Waele model

-
temperature at the surface and reaches the steady state temperature
distribution as 7 a.The thermal boundary layer thickness is seen
k
L
Nu
= thermal conductivity
= characteristic length
= Nusselt number
n = viscosity index in Ostwald-de Waele model
n-1.0 Pr = Prandtl number

'9 P
Re
T
t
U,
U
= parameter in Laplace transform
= generalized Reynolds number
= temperature
= time
= free stream velocity
= velocity component in streamwise direction
0 = velocity component in transverse direction
X = coordinate along streamwise direction
Y = coordinate normal to surface
r(i) = gamma function = Jt al-le%a
rdi) = incomplete gamma function = JE aj-le-ada
= nondimensional coordinate
0.001 0.01 0.1
T
1 10 10'
!
0
= nondimensional coordinate
= nondimensional temperature
Flgure 3. Transknl heal flux distribution for n = 1 and Pr = 5-1000. P = density

AlChE Journal (Vol. 28, No. 1) January, 1982 Page 59

9 = stream function in Laminar Boundary Layer Flows of Non-Newtonian Fluids Past Ex-
7 = nondimensional time ternal Surfaces,” AIChE ./., 6,312 (1960).
Gorla, R. S. R., “Unsteady Thermal Boundary Layers in Non-Newtonian
Fluids,” AIChE Symp. Ser., 76,264 (1980).
Subscripts Lee, S. Y. and W. F. Ames, “Similarity Solutions for Non-Newtonians
Fluids,’’AIChE ./., 12,700 (1966).
S = steady state conditions Luikov, A. V., “External Convective Mass Transfer in Non-Newtonian
W = conditions at the wall Fluid,” Int. 1.Heat Moss Transfer, 12,377 (1969).
m = conditions far away from the wall Metzner, A.B., “Heat Transfer in Non-Newtonian Fluids,” Adoances in
Heot Transfer, 2,357 (1965).
Shah, M. J.. E. E. Petersen, and A. Acrivos, “Heat Transfer from a Cylinder
Superscript to a Power Law Non-Newtonian Fluid,” AIChE J . , 8,542 (1962).
* = physical variables Wolf, C. J. and A. A. Szewczyk, “Laminar Heat Transfer to Power Law
Model Non-Newtonian Fluids from Arbitrary Cylinders,” Proceedings
of Third International Heat Transfer Conference, 1,388 (1966).
LITERATURE CITED
Manuscript received June 16,1980,revision received March 9, and accepted March
Acrivos, A,, M. J. Shah, and E. E. Petersen, “Momentum and Heat Transfer 18,1981.

Control System Synthesis Strategies

This paper is concerned with an important aspect of process control design- RAKESH GOVIND
synthesis of the control structure. Synthesis of control structures has long been
practiced by experienced control engineers, who relied on intuition, insight and Department of Chemical and Nuclear
judgment to pick a feasible solution from the vast number of alternatives that were Engineering
possible. This paper describes a systematic procedure to generate these alternatives Universityof Cincinnati
based on the cause-andeffect representation of the process. The final product is Cincinnati, OH 45221
a set of control schemes from which the final system may be selected or evolved. and
The work is significant in that it is the first attempt to apply non-numerical prob-
lem-solving techniques to the problem of synthesizing process control structures. G. J. POWERS
As such, it gives a new way of studying and teaching chemical process control. Department of Chemical Engineerlng
Carnegle-Melion University
Pittsburgh, PA 15231
SCOPE
Which variables should be measured, which inputs should 2. The propagation of control constraints through the process
be manipulated, and what links should be made between these flow diagram will generate the candidate control struc-
two sets? This is the essence of the synthesis of control structures tures.
in the chemical process industries. This problem is routinely 3. Evaluation of alternate control structures will depend pri-
solved by experienced engineers who have the ability to si- marily on: (a)the feasibility and simplicity of measuring and
multaneously consider: manipulating candidate control structure variables; and (b)
the steady-state interactions which occur between control
1. The economic, safety and reliability goals of a given pro-
substructures within the process. Detailed dynamic perfor-
cess mance needs to be considered only when the alternate con-
2. The steady-state and dynamic behavior of the complete trol structures have strong dynamic interactions or the dy-
process and of the pieces of equipment within it namic achievements of the control objective is question-
3. The interaction which might occur between control struc- able.
tures
4. The failure modes of the components within the process in- Hence, this approach has a strong component of steady-state
cluding the process operator control (Buckley, 1964; Shinskey, 1967, integrated with dynamic
5. Possible changes in the process to improve control considerations when it is indicated. These general concepts are
not new with us. Control engineers have used these ideas for
These engineers have evolved logical procedures for pro- decades. What we have done is to state the problem and solution
ceeding from loosely defined flowsheets and goals to wellde- method in the language of non-numerical problem-solving
fined piping and instrument diagrams (P&ID’s).Most of these (Newell, Shaw and Simon, 1959; Nilsson, 1971).In this language
procedures do not involve the use of detailed dynamic models the problem becomes one of using specific rules to generate a
of the process. How do they do it? This paper is our initial at- search tree of alternate control structures. The alternate struc-
tempt to capture at least part of the logic involved in trans- tures are evaluated using heuristics derived from both theoret-
forming steady-state flow diagrams into P&IDs. ical and practical considerations. The final product is a set of
Our approach is based on three main ideas: control schemes (say 5 to 10) from which the final system may
1. The models used in the synthesis of control systems must be be selected or evolved.
simple. To test these ideas, computer programs are being developed
which, while interacting with the control engineer, will generate
and evaluate alternate control systems. With these ideas and
computer programs, we hope to clarify the control synthesis
problem to generate other research projects.

AlChE Journal (Vol. 28, No. 1)